star-ga commited on
Commit
ffc42a3
·
verified ·
1 Parent(s): 11dd36e

docs: switch architecture to PNG embed (HF model cards render PNG reliably); tighten prose

Browse files
Files changed (1) hide show
  1. README.md +4 -4
README.md CHANGED
@@ -77,11 +77,11 @@ The cascade architecture (A gate + B specialist) is the result of **421 autonomo
77
 
78
  ## Architecture
79
 
80
- ![ClinicalMem BitNet b1.58 — A+B cascade architecture](architecture.svg)
 
 
81
 
82
- > *Source diagram (Mermaid): [`architecture.mmd`](architecture.mmd). Rendered to SVG with `@mermaid-js/mermaid-cli` v11.x and committed alongside the model so the architecture renders correctly on every viewer (HF model cards, GitHub, GitLab, NPM, etc.) without depending on a Mermaid runtime.*
83
-
84
- > **Source**: BitNet b1.58 architecture from Ma, Wang, Ma, et al. ([arXiv:2402.17764](https://arxiv.org/abs/2402.17764)). This is a clean-room Python implementation with **pure-integer Q16.16 fixed-point arithmetic** — no `torch` runtime dep, no GPU required. Training used PyTorch + Straight-Through Estimator on H200 SXM (RunPod).
85
 
86
  ---
87
 
 
77
 
78
  ## Architecture
79
 
80
+ <p align="center">
81
+ <img src="architecture.png" alt="ClinicalMem BitNet b1.58 — A+B cascade architecture" width="900">
82
+ </p>
83
 
84
+ Architecture from Ma, Wang, Ma, et al. ([arXiv:2402.17764](https://arxiv.org/abs/2402.17764)). Clean-room Python implementation with **pure-integer Q16.16 fixed-point arithmetic** no `torch` runtime dep, no GPU required. Training used PyTorch + Straight-Through Estimator on H200 SXM (RunPod). Diagram source: [`architecture.mmd`](architecture.mmd).
 
 
85
 
86
  ---
87